hexagon logo
Register Sign in
  • Replies 20 replies
  • Subscribers 222 subscribers
  • Views 217 views
  • Users 0 members are here
Options
Related

Hexagon wont help me, so now i ask you for help

pcdmisstudent
Hi guys & girls & friends & family.

I have a bigboy problem with my CMM, or well, rather a problem with the relationship between our 6 tips in our rack.
I've called hexagon support and they couldnt bring any help (also the good guy at support wasnt available today)
So now i want my friends (you guys) to help me! Cos i know some of you have the best knowledge n experience of this stuff!!
And... i cant wait for the other hexagon guy to help me, cos we have a big measuring error now..

So...I have two programs for checking our tips: a short & a long one.

Short one: I place a calibrated ring on our fixture plate. I measure & align to it with the master tip (with the rings top plane as Primary/Z-orgin and a circle with XY orgin).
Then it swaps to A0B0 for the next tip in our rack and measures the ring with a circle to evaluate Diameter, XY, Z(using a vector point on the face of ring) and also roundness to check for wear/dirt. I repeat this for all the other tips in the rack.
I get a big deviation in XY between the tips (0.05mm in X, 0.05mm in Y) and <0.01mm in Z. (I'll post the PDF report named "ring check")

The long check program uses the calibration sphere, where i use the same principle. Bolt the sphere to the fixture plate, Measure the sphere with mastertip A0B0 and align to it. Switch to another angle for the same tip and evaluate diameter/XYZ. Then swap to the next tip in the rack which measures the sphere and evaluates diameter/XYZ.
And I get the same big deviation in XY as in the ring check program. (Ive posted PDF report called "sphere check")

I have used these check programs for a long time and they've worked good so far.

So now....How do i get all the tips to show 0 in deviation (or at least <0.005mm per axis)?

Ive tried so far:

1. I ran our whole autocalibration program for all tips (which i also havent made any changes to) with good results, low STD, normal diameter deviation and the master XYZ is "perfect".
( ill post PDF report called "Autocalibration").

2. "Manually" calibrated the tips for the check program, i.e. I go into probe utlities for the first tip in our rack & calibrate A0B0 (master tip) saying "yes , sphere has moved". Then I go into the probe dialog for the 2nd tip in the rack and only calibrate A0B0 saying "no sphere hasnt moved". Ive done this for all tips in the check program with good results in the calibration results. But still bad results in the check program.

3. Ive reseted all tips to nominal and ran the autocalibration program again with good cal. results (But bad "check" results).

4. Ive inspected all tips for damage & cleaned them from dirt (also cleaned sphere/ring)

5. Played around with the calibration parameters for the tips, number of points & levels. And i acctually got sucess from this for the 2nd tip (Which you can see in the reports).
But i dont know why it worked, because it change anything for the other tips.

The only other thing i can think of that i havent tried is deleting the all probe files & build up new ones (If they became corrupt for some reason).

Anyone else has any ideas?
Just ask if u want more info, i can post code of the programs and show calibration settings etc...









  • Ok guys. I deleted the probe files and rebuilt them.
    I thoroughly cleaned all scales, tips, sphere & modules.
    Created an short autocalibration pogram with only a couple of angles and with cnc+cnc, 13 hits, 3 levels for each tip.

    After the calibration i ran the sphere check program, and they were all ALMOST good., some of my troublesome tips still gets halfbad deviation in XY (my disc & 0.5mm tip).
    Seems like i solved.....no........WE solved it
    I upload the results from the sphere check.
    Attached Files
  • Where are the CMM's? Is your environment/temperature in the room stable? We recently purchased the Hexagon Swift check artifact for monitoring CMM performance, and I have noticed some wild deviations when the temperature is fluctuating. I never even knew how bad until I put a temp gage in all of our inspection areas. The one area fluctuated 5°F in an hour at one point. And actually, the check it was failing was similar to your sphere check. It measures the sphere at T1A0B0 and aligns to it, then it comes down at all your 90° tips and re-measures. It then reports the worst offset. When the temperature was stable it passed the check. Your environment can have a massive impact on measurements if it is not stable.
  • in reply to pcdmisstudent
    I think the reason behind the weird measurements were a result from us recently getting an offline seat.
    And when i installed it and started programming it, i built our probe setup as it is. The probes on the online computer is acctually wrong. Theres a tesa-star head instead of our new HH-AS8-T5 (Didnt have to change em because they would be the same dimensions).

    Some tips were also a bit different from reality, our tip1/master is a diameter 2 ruby with lenght 30mm 1.5mm stem. It used to be a 2 ruby with a 20mm 1mm stem and then a 10mm long and 3mm diameter extension.

    It didnt make any difference in our measurements since they were nearly identical and corrected itself during calibration. But Im just guessing that when i imported the first program written offline (with new tips), the probe files on CMM computer became corrupt or something(?)
    not sure, just a theory.

    But it seems reasonable, because 09/25 our first offline program were run in the CMM, and the day after, the results from the check program were bad.
  • in reply to pcdmisstudent
    I do 24 hits and 3 levels for my calibration routines. I had issues with calibration deviations in the past and more hits have consistently given me better results. Calibrations take more time but the end justifies the means in my opinion.
  • in reply to pcdmisstudent
    What probe you have?
    I ve never had any problem with calibration deviation, this was some odd problem where calibration results were good, but practical measurement results were bad
    Ive touchtrigger probe tp200, and we used 7 points for years, with no issues. Ive then experimented abit, but it doesnt seem (according to me) to make a big difference between using 25pts as Hexagon recommends nor 7, or 9 etc..Wish i kinda had time to make a "real" study about it.

    Now i only picked 13 hits/3 levels for this new autocalibration because it seemed like a lot of people here on the forum use that number
  • in reply to Gagedude127
    Nice, luckily our cmm is in a climate controlled room



  • OK GUYS! IM BACK... to square one...

    I tried fixing this tip relationship problem yesterday. And eventually (as you may saw in my latest post) i seemed to finally fix the problem.

    I deleted all probe files, and built them from the ground up, using our acctual hardware (and not the hardware we used 10 years ago, tho the old hardware were pretty much the same..
    I calibrated the tips, got good values (yay) I checked the tips on the sphere and the ring & the deviation was only a few microns!! Slight smile YAY it measures correctly!

    Now that we we're all set and im ready to measure some parts. I welcome the operators kindly into the CMM room, proud of my self, showing them how good it measures now.
    So they ve been eager to measure all day, so we run our first part.
    And we have fixtures for all parts, so no manual alignment is needed.

    So the tip moves to the part (just a cylinder sitting on a v-block with the faces pointing in machine axis Y) . And it kinda goes to the centre to take a circle, but only in X/Y, but in Z it goes way down.. thinking the part is basically like 20-30mm lower than it should be..So i ran manual alignment again andt it measured the part correctly. (small yay)

    And it was like this for all other parts we tried to run, the machine always thinking the part was much lower in Z than it is.

    Anyone has any idea why this is? Maybe it has something to do with the machines zero? Or maybe its not a coincidence that before when my tips were showing a bad deviation in my check programs, it was only in bad in XY, but in Z it was just a few microns difference...and now its right in XY but not Z....hmmmh...

    Ok well whatever and i manually align the parts that were gonna be measured during night shift and went home...


    And now, this morning, what do u think happens?
    I run the daily check programs: AND ALL THE BIG DEVIATIONS ARE BACK! Even worse than before, like 0.1mm in X&Y!

    So today none of the tips relates to eachother again...for some reason...it just happend overnight..

    I need help again please
  • in reply to pcdmisstudent
    OK. step one..
    1. I shut off the machine and make zero it (Cos maybe me changing the head from tesa-star to hh-as8 makes a difference for the machine origo(?).
    2. I reset all tips to nominal again and check tips for damage (incase anything happend during night shift). No damage.
    3. I run a short autocalibration program, using just a few angles per tip. and it goes well, good calibration results.
    4. Run the Ring Check program again, checking all tips in A0B0. And now they re good. .....Besides the 2nd tip which shows 0.2mm dev in XYZ. (which is funny cos the 2nd tip was the only one relating to master before...)

    I see in probe utlities that that tip2 is D4 L40 D2 shaft.
    But in realiity it looks a bit different. Its D4 L10 D2 SHAFT + L30 D3 SHAFT.
    I changed so the tip is accuratly representing the acctual tip, and the results got better Slight smile

    Really i dont know why it would matter tho..
  • Question, do you have other sensors (optical or laser) in this CMM? Reason I ask is that one (or several) PC-DMIS versions had this bug where if you swapped sensors (from tactile to optical/laser/whatever) it did not retain the calibration. So, swapping from tactile -> optical/laser -> tactile would give you wrong results when measuring (due to the calibration not being retained bug).

    This was quite a nightmare because our autocal program measured tactile perfectly and gave splendid results prior to swapping sensors. The autocal went through tactile to optical to laser, and when we started measuring with the tactile after running the autocal program, all results were bad. Took quite some time to pin this down.

    Just a heads-up if you have several sensors in your autocal program.
  • in reply to vpt.se
    no sorry to disssapoint you vpt
    we arent as fancy as you guuys'
    we just use tp200, thats its
© 2024 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement